Stretching the imagination beyond muscle spindles – stretch‐sensitive mechanisms in arthropods

Abstract Much attention has been given to mammalian muscle spindles and their role in stretch‐mediated muscle proprioception. Recent studies, particularly, have sought to determine the molecular mediators of stretch‐evoked mechanotransduction, which these endings rely upon for functionality. Nonetheless, much about these endings remains unknown. Opportunities may be presented from consideration of extensive parallel research in stretch receptor mechanisms in arthropods. Such systems may provide a useful source of additional data and powerful tools for dissecting the complex systems of stretch transduction apparatus. At the least, such systems provide tractable exemplars of how organisms solve the problem of converting stretch stimuli to electrical output. Potentially, they may even provide molecular mechanisms and candidate molecular mediators of direct relevance to mammalian muscle spindles. Here we provide a brief overview of research on arthropod stretch receptors.


Introduction
To control posture and movement, animals with muscles sense muscle contraction and stretching via dedicated sensory organs, either directly or indirectly. Such muscle stretch receptor organs (SROs) are exemplified by the well-known mammalian muscle spindle (Bewick et al. 2005;Simon et al. 2010;Bewick & Banks, 2015). Invertebrate animals have a similar requirement for sensing muscle stretch, and such organisms represent a huge pool of biological diversity to be explored for mechanisms. As in other areas of neuroscience and physiology, invertebrate mechanisms provide a parallel route for study. Specifically, understanding the anatomy and physiology of stretch sensory organs in arthropods might be helpful for our appreciation of the mammalian muscle spindle.

Distribution and anatomy of stretch receptors in arthropods
The musculoskeletal systems of arthropods, particularly various crustaceans and insects, have a long history of study. SROs associated with muscles were first putatively identified anatomically in crustaceans by Alexandrowicz (1951). Between 1951 and1956, extensive analyses of 'muscle receptor organs' (MROs) were carried out in various crustacean species, rigorously characterising what turned out to be the stereotypical morphology of these putative receptors, being characteristically bipolar in shape with the two presumed sensory terminals oriented in parallel to the direction of the major longitudinal skeletal muscle fibres (Alexandrowicz, 1951(Alexandrowicz, , 1952(Alexandrowicz, , 1956. Subsequent studies in insects revealed similar simple putative stretch receptors (Finlayson & Loewenstein, 1958). These generally consist of single type II (i.e. non-ciliated) sensory neurons associated with strands of muscle or connective tissue (Br€ auning et al. 1986). Early studies identified a particular receptor that is generally termed the dorsal longitudinal stretch receptor or SRO. This SRO appears to be present in most thoracic and abdominal segments in larvae or adults of all insect orders investigated (Finlayson & Loewenstein, 1958;Osborne & Finlayson, 1962). The SRO consists of a sensory neuron with two sensory processes orientated longitudinally across the segment (Finlayson, 1976).
In fact, the presence of these bipolar dendritic neurons was first documented in the 18th century by the Dutch lawyer and amateur naturalist, Pierre Lyonet (Encyclopedia Brittanica, 2014), in superbly detailed drawings of the peripheral nervous system of the goat moth caterpillar, Cossus cossus L. (Lyonet, 1760). His drawings clearly show the presence in each segment of a pair of neurons with bipolar sensory processes, one on each side of the body wall, with their termini aligned across the longitudinal axis. Whilst the SRO is clearly homologous between diverse insects, there are differences in detail between the insect neurons, with bipolar dendrites, and those of crustaceans, in which more complex neuronal projections appear to insert into a longitudinal muscle capsule structure, and it is therefore unclear as to whether the insect SRO and crustacean MRO are completely homologous. Nevertheless, the presence of receptor termini arranged parallel to the main axis of stretch hints at a common function, and is also notably reminiscent of the arrangement of spindles within mammalian muscles.
In more recent years, these structures have been described in detail in the tobacco hornworm, Manduca sexta (L.), the larva of a type of hawkmoth (Tamarkin & Levine, 1996) and also in the fruit fly, Drosophila melanogaster Meigen (Schrader & Merritt, 2007). Studies in a range of insect orders show the dorsal longitudinal neurons to be associated along muscles. In Manduca the SRO consists of the bipolar neuron embedded within a thin 'receptor muscle' (Tamarkin & Levine, 1996). However, the equivalent neuron appears different in Drosophila. In surveys mapping the larval sensory nervous system of Drosophila, the equivalent cell to the SRO neuron has been termed the dorsal bipolar dendrite (dbd) neuron ( Fig. 1; Bodmer & Jan, 1987). This cell is in a similar location to other SROs and has a similar anatomy, with dendrites lying parallel to the main longitudinal muscle fibres. However, in this case the dendrites are not located along muscles directly, but instead are physically associated with a thin connective tissue strand and the distal tips of the sensory projections are attached to epidermal cells at the segment boundaries, similarly to the longitudinal muscle insertions themselves (Figs 1 and 2; Schrader & Merritt, 2007).
Much is known of the developmental origin of sensory neurons in Drosophila. The dbd neuron has several developmental features that are either shared with or distinct from other sensory neurons (Lai & Orgogozo, 2004). All sensory neurons originate from individual precursor cells derived from the ectoderm by the expression of a proneural transcription factor (Jarman, 2002). Unusually, the dbd neuron precursor cell requires the proneural gene, amos, which is otherwise mainly responsible only for olfactory sensory neurons (Huang et al. 2000;zur Lage et al. 2003). Moreover, the dbd precursor cell divides only once to give rise to the single dbd neuron and a cell of glial nature. This division pattern is quite distinct from that of all other sensory precursor cells, although comparative analysis has suggested that the dbd division pattern, a single division to give the neuron and its ensheathing sister glial cell, may in fact represent the ancestral state for sensory precursors (Lai & Orgogozo, 2004). During insect metamorphosis, many sensory neurons are completely replaced; however, the Drosophila larval dbd neurons have been shown to persist into the adult (Shepherd & Smith, 1996).

Stretch receptor function
Clearly, the SRO and related MRO are in a position to provide proprioceptive feedback during segmental muscle contraction and relaxation. In caterpillars, the SRO responds (as measured by firing rates) both tonically to slow stretch and phasically to rapid stretch (Osborne & Finlayson, 1965;Tamarkin & Levine, 1996;Simon & Trimmer, 2009). Moreover, SRO stimulation results in muscle stretch reflexes (Tamarkin & Levine, 1996). In species where these stretch receptors are  associated with a sensory 'receptor muscle', it appears that this allows modulation of stretch receptor response by efferent feedbacka situation somewhat reminiscent of the fusiform fibre workings of mammalian muscle spindles. In Manduca, the SRO neuron sends an axon to the central nervous system that makes direct synaptic connections with motor neurons (Tamarkin & Levine, 1996). It appears that motor neurons innervating muscles close to the receptor receive excitatory input, whereas muscles on the opposite side of the body are inhibited (Tamarkin & Levine, 1996).
Despite the functional evidence for the Manduca caterpillar SRO as a proprioceptor, a surprising observation is that surgical ablation of the SRO in one or several segments did not appear to result in a change in various movement behaviours (Simon & Trimmer, 2009). A role as an error detector is possible, or even some kind of developmental role, such as regulating larval moulting through responding to body size. The functional significance of this stretch receptor clearly requires more attention. In this context, the experimental tools available in Drosophila may suggest that this is one way forward. For example, genetic approaches could be used to ablate or silence all the dbd neurons in an individual, either developmentally or functionally, rather than surgically.

Physiology of arthropod stretch receptors
Electrophysiological studies have confirmed the stretch receptor action of these organs, and have often drawn parallels between their characteristics and those of vertebrate muscle spindles. Eyzaguirre & Kuffler (1955) demonstrated that a micrometre stretch stimulus applied to an intact nerve-muscle preparation from the crayfish Procambarus alleni (Faxon), Orconectes virilis (Hagen) and Homarus americanus Milne-Edwards evoked excitability of the afferent nerve. Notably, this study cites the methods of Katz, investigating stretch-evoked excitability of frog muscle spindles, as inspiration (Katz, 1950a,b). In particular, Eyzaguirre & Kuffler make special mention, on a couple of occasions, that their recordings bear similarities to the observations of Katz. Later examinations of the electrophysiological responses of stretch receptors have attempted to define the nature of the receptor potential generated in response to stretch. Work on the MRO of the crayfish Astacus astacus L. and Pacifastacus leniusculus (Dana) demonstrated the presence in these endings of stretch-activated tetrodotoxin-(TTX) and tetra-ethyl ammonium-(TEA) sensitive currents (Ottoson & Swerup, 1985a,b), suggesting the contribution of voltagegated Na + channels and voltage-gated K + channels, respectively (Kaila et al. 1987). A further Na + current component was detected that was activated by mechanical stimulation but not blocked by TTX. In addition, some Ca 2+ -dependence of the receptor potential that could be recorded in the sensory terminals was reported (Edwards et al. 1981).
What is particularly interesting about the receptor potentials of crayfish stretch receptors, as observed by, for example, Ottoson & Swerup (1982), is the similarity in waveform to analogous recordings made in rodent muscle spindles (Hunt et al. 1978). That is, recordings of receptor potentials in both crayfish stretch receptors and rodent muscle spindles exhibit an 'on' response to stretch, characterised by a large depolarisation response to dynamic stretch, an adaptive repolarisation to a 'hold potential' (persistent depolarised state) that is proportional to the amplitude of a static  stretch stimulus, and an 'off' response as stimuli are terminated, resulting in a return to resting membrane potential (Fig. 3). In the muscle spindle study, intracellular sharp electrode measurements of stretch-evoked receptor potentials, made from TTX-poisoned axons of Ia afferents, had also reported the ionic dependencies observed in the MRO: a strong TTX-insensitive mechanosensory Na + current, a smaller (~20%) contribution from Ca 2+ , and a TEA-dependent K + current (Hunt et al. 1978). The parallels are striking and, indeed, strongly point to the potential for some degree of mechanistic correlation between the two systems.
More recently, an analysis of the ionic dependencies of the dbd neuron of Drosophila has added further detail to support the anatomical evidence for its relationship to other arthropod stretch receptors (Nair et al. 2010). This study identified, via whole-cell voltage-clamp recordings, distinct 4-aminopyridine-sensitive and TEA-sensitive K + currents, and a Na + current blocked by TTX, as well as a voltage-activated Ca 2+ -current. Again, these distinctly resembled earlier recordings from stretch receptors of other species (Edwards et al. 1981;Ottoson & Swerup, 1982). Nair et al. (2010) concluded that their method might prove a valuable tool in finding further evidence to support the hypothesis that dbd neurons are likely stretch receptor candidates in Drosophila. Indeed, in our own work we have successfully adapted this protocol to confirm the existence of mechanically evoked potential responses in the dbd neuron dendrites of filleted larvae (Suslak et al. 2015).

Mediators of mechanotransduction
The nature of the mechanical stimulation leading to mechanotransduction is not known. In some insects it is proposed that free dendritic endings are squeezed during muscle or connective tissue stretch (Osborne & Finlayson, 1965). In the Drosophila dbd neuron, detailed structural analysis led to the proposal that mechanical stimulation results in dendritic membrane tension or in shearing forces between the dendrite and the surrounding glial cell (Schrader & Merritt, 2007).
Despite a wealth of anatomical and physiological knowledge, little is known of the molecular mediators of mechanotransduction in arthropod stretch receptors. In one approach, the application of mathematical modelling of these receptor types opens up a useful avenue of comparative study. Edwards et al. (1981) demonstrated that the Goldman-Hodgkin-Katz equation (Hodgkin & Katz, 1949) could be utilised in crayfish stretch receptors to estimate the ionic dependence of stretch-induced currents in the afferent ending. This method was further extended to model both the tension in the sensory endings as a Voigt element (spring and dashpot) and the role of a mechanosensory Na + current in stretch-mediated receptor depolarisation and receptor potential generation (Swerup & Rydqvist, 1996). Our own work has developed this in silico approach to reproduce both the stretch-and voltage-mediated currents observed by Ottoson & Swerup (1982). Significantly, this biophysical model can account for the complex receptor potential observed in both arthropod and mammalian stretch receptors (Suslak et al. 2011(Suslak et al. , 2015. In addition to this general approach, there is great potential in screening for mechanotransduction molecules using Drosophila, both to understand SRO function as a model proprioceptor and possibly also to inform the search for similar mediators in vertebrate muscle spindles. Not only do Drosophila possess a well-described example of an arthropod stretch receptor in the dbd neuron, but the sensory projections of these neurons have been shown to be amenable to patch-clamp electrophysiology in order to obtain direct measurements of receptor potential changes (Baines & Bate, Feline muscle spindle Crayfish stretch receptor B A Fig. 3 Receptor potentials of rodent muscle spindles and crayfish stretch receptors compared. (A) Receptor potential recorded from primary axon of a muscle spindle afferent in response to a stretch stimulus, shown below the trace (adapted from Hunt et al. (1978)). (1) Baseline; (2) peak of initial dynamic component; (3) peak of late dynamic; (4) post-dynamic minimum; (5) static maximum; (6) end static level; and (7) post-release minimum. (B) A receptor potential of crayfish SRO in response to stretch (adapted from Ottoson & Swerup, 1982). Resting membrane potential is indicated to the left of the trace. The crayfish trace has been annotated in a similar fashion to (A). In this interpretation, several features of the muscle spindle response are identifiable in the crayfish data. A notable exception is lack of post-release minimum (7). Other differences may be the result of the step-change stretch applied rather than a ramp stimulus.
1998; Nair et al. 2010). Such information has thus far been difficult to obtain in vertebrate muscle spindles. Indeed, we have recently demonstrated that receptor potentials of intact dbd neurons in response to a stretch stimulus can be recorded in situ (Suslak et al. 2015). The receptor potential profile of the dbd neuron appears strikingly similar to that of mammalian muscle spindle shown in Fig. 3A, including the post-stretch hyperpolarisation component (Suslak et al. 2015).
When combined with the powerful genetic toolbox available for Drosophila, this system has the potential to yield a wealth of data that could greatly advance our knowledge of mechanotransduction mechanisms. Indeed, several putative mechanosensory channels are known to be expressed in dbd neurons, including the Drosophila orthologues of Piezo and TRPA1 (Kim et al. 2010(Kim et al. , 2012Shen et al. 2011), and our recent work using a combination of pharmacology, genetics and electrophysiology has indicated that both channels play a role in dbd neuron receptor potential generation (Suslak et al. 2015). Interestingly, there is as yet no evidence for a role for ENaC channels in dbd neurons (Suslak et al. 2015), although they are strong candidates for involvement in mammalian muscle spindles (Simon et al. 2010). Another apparent difference with muscle spindles is that the ultrastructure of dbd neurons (Schrader & Merritt, 2007) appears to give no support for the action of an autogenic glutamate mechanism based on release of dendritic vesicles, as has been demonstrated for muscle spindles (Bewick et al. 2005;Bewick & Banks, 2015).
Currently, then, study of stretch receptors of arthropods presents a useful avenue for addressing the question of what may mediate stretch-evoked responses in mechanoreceptors. Moving forward, the use of the model organism D. melanogaster may provide much new insight into what constitutes a mechanosensory mechanism. Such insight may provide an intriguing and useful comparison to the mammalian muscle spindle.